Increasingly, users use their mobile client devices to capture and share images on different network platforms (e.g., social media network sites). The mobile client devices have limited computing resources (e.g., lower power hardware processors, limited memory) and cannot efficiently execute complex image processing processes, such as execute a convolutional neural network to perform style transfers.